Martin Karplus of Harvard University in the U.S. will integrate informatics and artificial intelligence approaches to design stable, synthetic antigens based on the viral hemagglutinin (HA) protein to be used as a universal influenza vaccine. Seasonal influenza causes substantial morbidity worldwide, and the development of a universal vaccine is a global health priority. The current HA-based vaccines do not provide broad coverage against multiple strains, and must be administered annually because of the high mutability of the virus. To address this, they will use a new approach to develop vaccines that elicit broadly-neutralizing antibodies effective against many different influenza strains, including avian, swine, and human varieties. First, they will assemble the publicly-available DNA sequences of the HA gene of avian, swine, and human influenza A strains, and use homology modeling with existing protein structures as templates to create a library of antigens that together are predicted to both maintain the immune system’s memory of influenza while enabling a rapid immune response to future seasonal or pandemic strains. The antigen library will be optimized by performing affinity maturation simulations initiated from known germline precursors of broadly-neutralizing antibodies, and combining the results with machine learning, to predict which antigen cocktails will produce antibodies with the greatest breadth of reactivity. The most promising of these immunization cocktails will be efficacy-tested in small animal models.

Peter Kwong of the Vaccine Research Center, National Institute of Allergy and Infectious Diseases in the U.S. will use an informatics-based approach to identify influenza virus epitopes that are especially suited to induce a strong and broad immune response, being conserved, accessible, and with a specific structural flexibility, and develop them as vaccine targets. Influenza is highly contagious and can cause severe illness, particularly among the young, elderly, and those with pre-existing conditions. Current vaccines are only partially effective, protect against single strains, and are generally ineffective against pandemic strains. Starting with three exposed influenza viral proteins, hemagglutinin (HA), neuraminidase (NA), and matrix protein 2 ectodomain (M2e), they will identify and rank candidate epitopes by determining their structural flexibility using all-atom molecular dynamic simulations, and by evaluating their conservation and surface accessibility, to promote recognition by the immune system and the generation of antibodies that target different viral strains. Candidate epitopes will be tested for antigenicity in vitro, and the best used to immunize mice to characterize the antibody response. After optimizing the vaccine regimen, they will ultimately evaluate the ability of the candidate vaccines to provide protection against diverse influenza strains in multiple animal models.

Dr. Yoshihiro Kawaoka of the University of Tokyo in Japan will develop broadly effective influenza vaccines by mixing together epitopes of conserved fragments of the viral hemagglutinin (HA) protein, which only elicit a weak immune response, together with millions of different, non-naturally occurring fragments that elicit a strong immune response, to induce broadly cross-reacting antibodies. Influenza is of world-wide concern severely impacting public health and the global economy. Tens of millions of reported cases result in tens of thousands of deaths annually in the U.S. alone, and the rapid spread of the virus between countries causes epidemic or pandemic outbreaks. Traditional vaccines are directed towards selected epitopes in the head region of the viral HA protein because they elicit a strong immune response. However, these regions are frequently mutated, rendering the vaccines useless. Vaccines directed towards more conserved epitopes only elicit a weak immune response, but this can be strengthened using HA proteins previously unseen by the immune system. They will produce a library of millions of HA epitopes that contain artificial mutations in the immune-dominant regions while preserving the conserved regions. This should focus the production of highly reactive antibodies against the conserved HA epitopes, which will eliminate a wider range of influenza strains. They will test this using single and repeat immunizations of different mixes in ferrets. Once optimized, the vaccination strategy will be tested in ferrets pre-exposed to influenza virus to mimic the human situation. The result will be a single vaccination that protects against a wider range of influenza strains than traditional vaccines.

Alice McHardy of the Helmholtz Centre for Infection Research in Germany will use a computational approach to engineer a more stable, neuraminidase (NA)-like antigen for use in influenza vaccines to increase both the duration and the breadth of protection against multiple influenza strains. Seasonal influenza viruses are constantly mutating, and current vaccines designed to target the variable, but strongly immunogenic, surface antigen, hemagglutinin (HA), must be frequently replaced. Vaccines designed to target another immunogenic antigen, neuraminidase (NA), fail to elicit a strong immune response likely due to the instability of the NA protein during the manufacturing process. They will combine genetic, structural, and evolutionary data to design an NA-like antigen with epitopes conserved across at least two influenza subtypes to provide protection against a broader range of influenza strains. The protein will be optimized for stability and immunogenicity, and then tested in mouse and ferret influenza infection models. The antigen will be co-formulated with c-di-AMP (known to improve responsiveness to vaccines in vulnerable populations including infants and the elderly) as the adjuvant. The expected result of this project is an antigen that provides increased duration and breadth of immune protection against multiple strains of influenza.